API Integration

Push data programmatisk til din AI SmartTalk vidensbase ved hjælp af vores REST API. Ideel til tilpassede integrationer, automatiserede pipelines og enhver datakilde, der ikke dækkes af vores native connectors.

---

Oversigt

API-integrationen gør det muligt for dig at:

• Push dokumenter direkte til din vidensbase
• Opdatere indhold programmatisk
• Slette forældede poster
• Bygge tilpassede datapipelines
• Integrere med ethvert system, der kan foretage HTTP-anmodninger

---

Forudsætninger

Før du begynder, skal du sikre dig, at du har:

• En aktiv AI SmartTalk-konto
• API-adgang aktiveret (tjek din plan)
• Grundlæggende viden om REST APIs
• Et værktøj til at foretage HTTP-anmodninger (curl, Postman eller din applikationskode)

---

Få Dine API-legitimationsoplysninger

Trin 1: Adgang til API-indstillinger

1. Log ind på din AI SmartTalk-konto
2. Naviger til Indstillinger → Integrationer
3. Find API og klik på Konfigurer

Trin 2: Generer API-token

1. Klik på Generer nyt token
2. Kopier dit Chat Model ID og API-token
3. Opbevar disse sikkert—token vises kun én gang!

⚠️ Sikkerhedsadvarsel: Udsæt aldrig dit API-token i klient-side kode eller offentlige repositories.

---

API Endpoints

Base URL

https://api.aismarttalk.tech/v1

Autentificering

Alle anmodninger kræver dit API-token i headeren:

Authorization: Bearer YOUR_API_TOKEN

---

API Endpoints

Base URL

https://api.aismarttalk.tech/v1

Authentication

Alle anmodninger kræver din API-token i headeren:

Authorization: Bearer YOUR_API_TOKEN

---

Import Documents

Endpoint

POST /documents/import

Request Body

{
  "chatModelId": "your-chat-model-id",
  "documents": [
    {
      "title": "Produktdokumentation",
      "content": "Det fulde indhold af dit dokument går her...",
      "url": "https://example.com/docs/product",
      "metadata": {
        "category": "dokumentation",
        "language": "en"
      }
    }
  ]
}

Parameters

Field	Type	Required	Description
chatModelId	string	✅	Dit unikke chatmodel-id
documents	array	✅	Array af dokumentobjekter
documents[].title	string	✅	Dokumenttitel til identifikation
documents[].content	string	✅	Fuld tekstindhold
documents[].url	string	❌	Kilde-URL (til reference)
documents[].metadata	object	❌	Brugerdefinerede nøgle-værdi-par

Response

{
  "success": true,
  "imported": 1,
  "documents": [
    {
      "id": "doc_abc123",
      "title": "Produktdokumentation",
      "status": "behandler"
    }
  ]
}

Example: cURL

curl -X POST https://api.aismarttalk.tech/v1/documents/import \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "chatModelId": "your-chat-model-id",
    "documents": [
      {
        "title": "Kom godt i gang guide",
        "content": "Velkommen til vores platform. Her er hvordan du kommer i gang...",
        "url": "https://docs.example.com/getting-started"
      }
    ]
  }'

---

Query Documents

Stil spørgsmål til din vidensbase programmatisk.

Endpoint

POST /chat/query

Request Body

{
  "chatModelId": "your-chat-model-id",
  "query": "Hvordan nulstiller jeg min adgangskode?",
  "options": {
    "maxTokens": 500,
    "temperature": 0.7
  }
}

Response

{
  "success": true,
  "response": "For at nulstille din adgangskode, gå til Indstillinger > Sikkerhed > Ændre adgangskode...",
  "sources": [
    {
      "documentId": "doc_abc123",
      "title": "Sikkerhedsguide",
      "relevance": 0.95
    }
  ]
}

---

Retrieve Documents

Hent dokumenter, der matcher en forespørgsel (uden AI-respons).

Endpoint

POST /documents/search

Request Body

{
  "chatModelId": "your-chat-model-id",
  "query": "adgangskode sikkerhed",
  "limit": 10
}

Response

{
  "success": true,
  "documents": [
    {
      "id": "doc_abc123",
      "title": "Bedste praksis for sikkerhed",
      "content": "...",
      "relevance": 0.92
    }
  ]
}

---

Kodeeksempler

Python

import requests

API_TOKEN = "your-api-token"
CHAT_MODEL_ID = "your-chat-model-id"

def import_document(title: str, content: str, url: str = None):
    response = requests.post(
        "https://api.aismarttalk.tech/v1/documents/import",
        headers={
            "Authorization": f"Bearer {API_TOKEN}",
            "Content-Type": "application/json"
        },
        json={
            "chatModelId": CHAT_MODEL_ID,
            "documents": [{
                "title": title,
                "content": content,
                "url": url
            }]
        }
    )
    return response.json()

# Importer et dokument
result = import_document(
    title="FAQ: Forsendelse",
    content="Vi tilbyder gratis forsendelse på ordrer over $50...",
    url="https://shop.example.com/faq/shipping"
)
print(result)

JavaScript / Node.js

const API_TOKEN = 'your-api-token';
const CHAT_MODEL_ID = 'your-chat-model-id';

async function importDocument(title, content, url = null) {
  const response = await fetch('https://api.aismarttalk.tech/v1/documents/import', {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${API_TOKEN}`,
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      chatModelId: CHAT_MODEL_ID,
      documents: [{
        title,
        content,
        url
      }]
    })
  });
  return response.json();
}

// Importer et dokument
importDocument(
  'FAQ: Returneringer',
  'Du kan returnere varer inden for 30 dage efter køb...',
  'https://shop.example.com/faq/returns'
).then(console.log);

PHP

<?php
$apiToken = 'your-api-token';
$chatModelId = 'your-chat-model-id';

$data = [
    'chatModelId' => $chatModelId,
    'documents' => [
        [
            'title' => 'Produktspecifikationer',
            'content' => 'Vores widget har følgende specifikationer...',
            'url' => 'https://example.com/products/widget'
        ]
    ]
];

$ch = curl_init('https://api.aismarttalk.tech/v1/documents/import');
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_POST, true);
curl_setopt($ch, CURLOPT_HTTPHEADER, [
    'Authorization: Bearer ' . $apiToken,
    'Content-Type: application/json'
]);
curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($data));

$response = curl_exec($ch);
curl_close($ch);

print_r(json_decode($response, true));

---

Anvendelsessager

Tilpasset CMS-integration

Synk indhold fra et proprietært CMS:

• Hook ind i CMS publiceringsbegivenheder
• Push nyt/opdateret indhold til AI SmartTalk
• Fjern slettet indhold

Datapipeline

Importer fra datalagre:

• Eksporter relevante data til JSON
• Batchimport via API
• Planlæg regelmæssige opdateringer

E-handelsprodukter

Synk produktdata fra tilpassede systemer:

• Produktbeskrivelser
• Specifikationer
• Prisinformation

Interne systemer

Forbind interne værktøjer, der ikke understøttes nativt:

• Tilpassede wikis
• Arv databaser
• Proprietære applikationer

---

Rate Limits

Endpoint	Rate Limit
Document Import	100 requests/minute
Query	60 requests/minute
Search	60 requests/minute

Note: Rate limits varierer afhængigt af planen. Kontakt support for højere grænser.

---

Error Handling

Error Response Format

{
  "success": false,
  "error": {
    "code": "INVALID_TOKEN",
    "message": "The provided API token is invalid or expired"
  }
}

Common Error Codes

Code	Description	Solution
INVALID_TOKEN	Bad or expired token	Regenerate API token
INVALID_MODEL_ID	Unknown chat model ID	Check your Chat Model ID
RATE_LIMITED	Too many requests	Implement backoff, retry later
INVALID_REQUEST	Malformed request body	Check JSON structure
DOCUMENT_TOO_LARGE	Content exceeds limit	Split into smaller documents
QUOTA_EXCEEDED	Plan limits reached	Upgrade or contact support

---

Troubleshooting

Authentication Fails

Issue	Solution
401 Unauthorized	Check token is correct and active
Token not working	Regenerate token in settings
Expired token	Tokens don't expire, but can be revoked

Import Issues

Issue	Solution
Empty response	Check Content-Type is application/json
Document not appearing	Wait for processing; check Knowledge section
Partial import	Some documents may have validation errors

Performance Issues

Issue	Solution
Slow imports	Batch documents (max 100 per request)
Timeouts	Reduce batch size, retry with backoff
Rate limited	Implement exponential backoff

---

Best Practices

1. Batch imports: Send multiple documents per request (up to 100)
2. Unique titles: Use descriptive, unique titles for each document
3. Structured content: Well-formatted content improves AI responses
4. Metadata tagging: Use metadata for categorization and filtering
5. Secure tokens: Store tokens in environment variables
6. Handle errors: Implement retry logic with exponential backoff
7. Monitor usage: Track API calls against your plan limits

---

Related Documentation

• Integrations Overview
• Full API Documentation
• Knowledge Base Management
• SmartFlow Webhooks — Receive events